Here is the issue. I ran a logistic regression a time variable (time to surgery) had a trend of increased infection. Great right. Done? No. This is where SPSS gets tricky and why I might just abort back to SAS like I do sometimes. But lets see.
I ran a graph for cumulative infection count versus time. It has a big slope initially and then flattens.
The cumulative count functions for SPSS seem to be nonexistent. That makes zero cents. Why would you graph something only? Making a variable may be useful for analyses.
Essentially, I want to do a couple things.
1. I want to run a regression with a cumulative count or percentage or whatever. This makes the outcome variable a count that I can graph, I can regress etc.
2. I want to go back and do log regression analyses by cutpoints/ties/knots whatever you want to call them and do something curvilinear with count variable.
Is this possible?
I've spent hours dinking with SPSS and this should be straight forward. if you can graph something as a cumulative sum and the SPSS also has a curve estimator, why is this program so limited with dichotomous variables? Think what I'm requesting to do is pretty straight forward for me. Fact matter is I can probably do this in SAS by now is kinda sad.
I ran a graph for cumulative infection count versus time. It has a big slope initially and then flattens.
The cumulative count functions for SPSS seem to be nonexistent. That makes zero cents. Why would you graph something only? Making a variable may be useful for analyses.
Essentially, I want to do a couple things.
1. I want to run a regression with a cumulative count or percentage or whatever. This makes the outcome variable a count that I can graph, I can regress etc.
2. I want to go back and do log regression analyses by cutpoints/ties/knots whatever you want to call them and do something curvilinear with count variable.
Is this possible?
I've spent hours dinking with SPSS and this should be straight forward. if you can graph something as a cumulative sum and the SPSS also has a curve estimator, why is this program so limited with dichotomous variables? Think what I'm requesting to do is pretty straight forward for me. Fact matter is I can probably do this in SAS by now is kinda sad.
Attachments

107.6 KB Views: 0